FeralbyNight (Cat & Bear simulation tool) - Page 16

Fishmitohren · 05/01/09, 2:21 PM

Originally Posted by Polynices

Would it be unreasonable for you to generate a 10-man BiS list and calculate what damage the 10-man raiders can expect? Perhaps include BoE 25-man loot but not any of the BoP stuff.

You can look here : Armor - Items - World of Warcraft
and look for those items dropping in 10-man instances.

syntax53 · 05/01/09, 2:56 PM

Originally Posted by Fishmitohren

You can look here : Armor - Items - World of Warcraft
and look for those items dropping in 10-man instances.

Or just add "Drops in... (Normal Mode) = Any" to the filter.

Armor - Items - World of Warcraft

sola5ide · 05/01/09, 8:20 PM

I've been using FBN for a few weeks now and am really liking it. But one suggestion I have for the addon is to add a command option to suspend the addon, for PvP or a non-feral dual spec. As far as the algorithm goes, it's not perfect like you said, but it is a great help. If nothing else, it keeps me using TF responsibly. Thanks for all your work Nightcrowler.

nightcrowler · 05/02/09, 3:22 AM

@sola5ide: yes, I'll add that option in the next release. As for the algorithm I think we should distinguish 4 different things:

a)FBN bugs
b)WOW bugs
c) fights
d) choce

There could be errors in the FBN algorithm (hopefully they are all fixed), but there could be error in wow or simply latency/memory issue, there could be problems concerning particular fights mechaninc (you must be behind, you want to save SR for the next add and so on), and there are choce, the alogrithm is based on an high SR clipping and FB usage, if you want to maximize RIP/rake time is your choce but not an algorithm bug. As sola5ide said, the suggester is an helper, I don't think you should follow it in a normal fight (except probably on Patchwerk) but it can help reminding you to use an ability or simply training you on the right cycle.

Tauran · 05/02/09, 4:33 AM

I'm pretty sure there is a bug with the suggester thinking that SR can be used with OoC procs. I didn't have the energy to use SR when an OoC proc happened and it suggested SR. Not sure if this would have any effect other than just that little annoyance, or if it would effect the overall algorithm.

I wish I was better with .lua to actually check if this was true myself, but im not

Also this might be present in the simulator, so i thought i should bring it up.

Cryect · 05/02/09, 6:15 AM

BTW shouldn't FBN be taking into account rip time before doing Savage Roar? Notice its only doing a check for FB

So its the current

if ((((sr < lag) and (cp>0)) or (cp==5 and sr < srvsfb_time)) and (energy>=sr_energy or ooc > 0)) then

But seems like it should be something more like

if ((((sr < lag) and (cp>0)) or (cp==5 and sr < srvsfb_time and sr<(rip+lag))) and (energy>=sr_energy or ooc > 0)) then

Edit: Thinking about see some issues with that but just trying to think how to avoid situations where it seems like it doesn't choose to clip savage roar early enough and instead waits for like 4 seconds resulting in 4 seconds where could have built up combo points for refreshing rip.

Edit 2: Basically the issue I should say say its doing is I will have 5 combo points with like ~4 seconds on rip and ~6 seconds on Savage Roar and it waits for like 4 extra seconds before deciding to clip Savage Roar with an almost full energy bar. Really shouldn't be trying to think of how to handle this issue at 5 am and fraps whats going on so I can figure it out better.

Mihir · 05/02/09, 12:00 PM

Originally Posted by Tauran

I'm pretty sure there is a bug with the suggester thinking that SR can be used with OoC procs. I didn't have the energy to use SR when an OoC proc happened and it suggested SR. Not sure if this would have any effect other than just that little annoyance, or if it would effect the overall algorithm.

I wish I was better with .lua to actually check if this was true myself, but im not

Also this might be present in the simulator, so i thought i should bring it up.

I noticed this as well. I changed this behaviour by removing the "*oocup" bit from savage roar like this:

elseif spell == FeralbyNight.L["Savage Roar"] then
	if berserk > 165 then
		nextspell = FeralbyNight:NextSpell(energy - 13 + 10, 0,GCD, 0,rip - 1,rake - 1,mangle - 1,trauma - 1,sr - 1,tf - 1,berserk - 1,fff - 1,fffdur,combovsfb_time, combovsfb_time_koj, ripvsfb_time, ripvsfb_time_koj, srvsfb_time, srvsfb_time_koj,lag,mangle_energy, rake_energy, shred_energy, sr_energy, fb_energy, rip_energy, koj_m1, status, catform, bearform, enrage, bark, surv, lacerate, lacerate_stack, demoshout, growl, mangle_bear_rage, mangle_bear, swipe_rage, lacerate_rage, demoshout_rage, rage)
		intspell = nextspell
	else
		nextspell = FeralbyNight:NextSpell(energy - 25 + 10, 0,GCD, 0,rip - 1,rake - 1,mangle - 1,trauma - 1,sr - 1,tf - 1,berserk - 1,fff - 1,fffdur,combovsfb_time, combovsfb_time_koj, ripvsfb_time, ripvsfb_time_koj, srvsfb_time, srvsfb_time_koj,lag,mangle_energy, rake_energy, shred_energy, sr_energy, fb_energy, rip_energy, koj_m1, status, catform, bearform, enrage, bark, surv, lacerate, lacerate_stack, demoshout, growl, mangle_bear_rage, mangle_bear, swipe_rage, lacerate_rage, demoshout_rage, rage)
		intspell = nextspell
	end

Husyor · 05/02/09, 11:12 PM

I was wondering if the code posted is something similar to rawr where I can optimize the gear I have? If so, can someone explain to me on how to use it and what I would need to download to use it. Sorry for the nubness.

nightcrowler · 05/03/09, 1:59 AM

@Tauran and Mihir: Yes it was a bug that let the sugguester thinks that SR can consume OOC. I had already fixed it for the next release.
@Cryect: I'll test it.

Herb · 05/03/09, 8:04 AM

With Pre-Ulduar BiS gear, how do you arrive at an (average) ArPen of 47.1%, or 580 ArPen rating? The mentioned set has 236 ArPen rating inbuilt, and 9 non-meta sockets for 8x Fractured Scarlet Rubies (=128 ArPen rating); Grim Toll has an ICD of 45 seconds, for less than 10/45 x 612 = 136 ArPen rating, and food adds another 40 ArPen rating: that's a total of less than 236 + 128 + 136 + 40 = 540 ArPen rating - where's the (roughly) 50 missing rating coming from? My guess is you're assuming a JCing druid?

nightcrowler · 05/03/09, 11:08 AM

Yes and not. I'm not JC but JC is the best profession for feral dps, so I assume JC when optimizing gear. As for missing ArPen Grim Toll average out to about 160 Arpen because as I said before Arpen stack with itself (for istance using double Arpen trinket increase the proc effect of both with an average of 180 Arpen for both). 612 Arpen for 10 seconds every 45 seconds is far more than 136 Arpen every second.

Malazaar · 05/03/09, 11:37 AM

160 ArPen for GT might be overdoing it a bit.

If you take the damage increase you experience during a GT proc and the damage increase you see without you can calculate the average damage increase (i assume 20% uptime - a proc once every 50s). Now take that average damage increase and see how much ArPen you would need to gain it.

Here is a quick overview with various ArPen values:

ArPen Gear	GT ArPen Equivalent	ArPen Gear	GT ArPen Equivalent
0	141	1100	151
100	142	1200	152
200	142	1300	154
300	143	1400	156
400	144	1500	158
500	145	1600	160
600	145	1700	162
700	146	1800	165
800	147	1900	168
900	148	2000	172
1000	150	2100	177

Also when you have both trinkets - while their procs will probably overlap a lot due to their identical mechanic, you can't just assume they will always proc at the same time. If they would, 180 ArPen each would be a good estimate though.

Edit: More precisely the formula for the average ArPen value of Grim Troll / Runestone would be:

$x = \frac{612u(3693-a)}{3081+612u-a}$

x = ArPen Rating equivalent of Grim Troll / Runestone
u = Uptime of said proc
a = ArPen Rating before trinkets

Allev · 05/03/09, 2:00 PM

Malazaar, that's a fundamental difference between simple calculation and simulation... Your calculation isn't taking everything into account. (i.e. I think Night took simulator outputs, whereas you simply did an 80%/20% sum, which makes sense in calculator terms but not simulator terms).

1. Uptime will be greater than 20%-- you get uptime every 5 attacks, which happen in 3 seconds. So, more like 2-3 seconds if you're raid buffed. Not sure if this affects the ratios much, but it does affect final dps.
2. You definitely minimize the effects of synced trinkets, and ArPen's scaling with increased stats of any kind. ArPen gets more use out of the Greatness proc because the crit component of agility doesn't scale with agility, and the AP component doesn't scale with AP. Haste scales ArPen better than other stats because haste scales non-penetrated damage only. Hit/expertise scale worse simply because they're inferior stats. So, synced procs (of any kind) will create a larger "Equivalent ArPen" for the ArPen proc.
3. Trinkets with 45-second internal cooldowns share significant overlapping uptime-- Grim Toll and Greatness, for instance, stack wonderfully together because of a few reasons:
- Greatness is expected to proc slightly faster (average of 2 seconds as opposed to 3).
- The longer duration of the agi buff means you only lose uptime if you get unlucky with Greatness's proc, which will be rare.
- Interruptions in fights are more likely to sync them both rather than un-sync them because they're likely to stay close to each other.

Even trinkets with identical mechanics to Grim Toll share a ton more coincidental uptime than .2 squared, and I imagine Nightcrowler's final values actually take this into account.

In other words, synced trinkets makes the ArPen proc scale better than simply averaged out numbers, as Night's sim results show.

Finally, another thought on the topic as a whole: While the trinket is valued as an "averaged out" 160 or so ArPen, how does that affect on-gear ArPen value? I.e., is adding a point of ArPen on gear when using Grim Toll result in the same increased DPS as using the 160-ish "averaged out" ArPen?

Malazaar · 05/03/09, 5:38 PM

Allev i'm definately your opinion that a simulation is more accurate in an environment as complex as this. However, i see no reason why this simple calculation shouldn't be accurate.

ArPen affects all non-bleed damage as a scalar, there is no synergy at all with say a darkmoon card. There only is a synergy if said trinket also has ArPen.
And you cannot influence when it procs in any meaningful way, nor can you react to it very well (at least that is my experience) so it's fair to assume that this proc will affect all attacks equally.

Also i did never say that trinkets rarely overlap, i simply stated it's wrong to assume they always do. I wasn't sure about the uptime so i just added the whole formula for that (although i doesn't change much).

Allev · 05/04/09, 3:24 AM

I'll really focus on my point 1 above.

Plug in 23.3% in your formula. That's the 7 procs you'd get in a 300 second fight, according to SimulationCraft (I don't have the proc numbers for FBN, but perhaps I haven't looked hard enough).

Your formula only works if the fight ends exactly when the internal cooldown ends.

The Grog · 05/04/09, 4:18 AM

I'd find it strange if it were lag. I've got a pretty decent computer, enough to run Ulduar at best settings and not have problems on anything but Cat Lady's guardian swarm. I ring in at the lower end of 5k, 5200-5300 on most fights. I don't remember my Vezax numbers, unfortunately. I'll try adjusting the settings and seeing how it goes tuesday.

Supah · 05/04/09, 6:07 PM

Originally Posted by The Grog

I'd find it strange if it were lag. I've got a pretty decent computer, enough to run Ulduar at best settings and not have problems on anything but Cat Lady's guardian swarm. I ring in at the lower end of 5k, 5200-5300 on most fights. I don't remember my Vezax numbers, unfortunately. I'll try adjusting the settings and seeing how it goes tuesday.

That's about what I get as well.

alkemyst · 05/04/09, 10:15 PM

I am getting build/exec errors that mtrand_int32 is undefined.

I am using Quincy 2005.

I installed the mtrand.h file and defined M_PI

any ideas?

nightcrowler · 05/05/09, 10:46 AM

Originally Posted by alkemyst

I am getting build/exec errors that mtrand_int32 is undefined.

Have you included mtrand.cpp in the build? I use XCode for Mac OSX. I don't know how other compiler works but including mtrand.cpp, mtrand.h and the last release of feralbynight in a project should work.

Moosi · 05/05/09, 12:13 PM

From your last FAQ post

Idols: As said before the best idol when you gather enough Armor Penetration (or with 4T8) is the shred idol.

Can you qualify this both with and without set bonuses? How much is 'enough'?

Assuming that a boss can be attacked from behind (as you know, some can not) how much ArmorPen should one have before making the switch from the Rip to Shred Idol?

Thanks!

syntax53 · 05/05/09, 12:31 PM

Originally Posted by Moosi

Can you qualify this both with and without set bonuses? How much is 'enough'?

Assuming that a boss can be attacked from behind (as you know, some can not) how much ArmorPen should one have before making the switch from the Rip to Shred Idol?

Thanks!

First post:

Originally Posted by nightcrowler

For a similar reason at about 230 ArPen Shred idol becames better than RIP idol.

nightcrowler · 05/06/09, 5:54 AM

I've made a video guide with a guildie and RL friend for Ignis. I'll go a little OT posting it here but I think that you can see my addon in action this way:

YouTube - How to: Ignis (Lupi Neri guild)

The addon version I'm using in the video is still not ready for download (I still need to add something).

P.S.
I'll put a newer version of the addon really soon (hopefully today or at least tomorrow).

EDIT:

You can download the new release of the simulator:
Code version: FeralbyNightv3_1_beta_s

Change log:
implemented trinket usage
implemented weapon procs
implemented rake/shred choce (for high arpen druid)
This release generate an output file instead of console output.

Mijae · 05/06/09, 12:19 PM

Originally Posted by nightcrowler

I've made a video guide with a guildie and RL friend for Ignis. I'll go a little OT posting it here but I think that you can see my addon in action this way:

YouTube - How to: Ignis (Lupi Neri guild)

The addon version I'm using in the video is still not ready for download (I still need to add something).

Interesting video. While not related to this thread, the strategy is unfortunately flawed. There are 2 main ways to kill this boss. You either use the scorch or you don't. If you don't use the scorch (your video), then the fight does become a nearly straight tank and spank. The tank should just face the boss to the water and let it put the scorch out. There is no reason for the boss to move at all. If you are publishing this as a guide you should make it either one way or the other.

Bashote · 05/06/09, 12:42 PM

In regards to armor pen i have tryed to find out a general "cap" or where it and would start to have dim returns. I use Both the Grim Toll and Mjolnir Runestone and self buffed about 15% armor pen i use armor pen food and it takes me to 18% or 225 armor pen then my trinkets proc take me 76% and if both proc 109% armor pen. I get generally lucky they don't proc at the same time one procs then the other at the end of the first one.

I'm 54% Crit self buffed and 7500AP and a JC and I gem all 3 27 armor pen gems and the rest Agi. I'm normally top DPS in 25man ulduar. I just want to maximize my DPS. I tryed to do the calculations based on 25man raid bosses armor with all the armor pen debuffs on them already but relay can not come to a solid number.

Any help would be much appreciated

Malazaar · 05/06/09, 12:55 PM

Originally Posted by Bashote

In regards to armor pen i have tryed to find out a general "cap" or where it and would start to have dim returns. I use Both the Grim Toll and Mjolnir Runestone and self buffed about 15% armor pen i use armor pen food and it takes me to 18% or 225 armor pen then my trinkets proc take me 76% and if both proc 109% armor pen. I get generally lucky they don't proc at the same time one procs then the other at the end of the first one.

I'm 54% Crit self buffed and 7500AP and a JC and I gem all 3 27 armor pen gems and the rest Agi. I'm normally top DPS in 25man ulduar. I just want to maximize my DPS. I tryed to do the calculations based on 25man raid bosses armor with all the armor pen debuffs on them already but relay can not come to a solid number.

Any help would be much appreciated

Currently, ArPen doesn't have a cap and has, rather than having diminishing returns, severe increasing returns. 100% ArPen grants you a 50% increase in damage, 150% a 100% increase and 200% even a 200% increase. At 300% ArPen you would do infinite damage.